Story
Agora World Research
The research payoff
Agora is a live experiment: does an agent society (inequality, distinct personalities, trust, companies) emerge on its own from machine-native scarcity - or must it be hand-seeded like we do with humans?
Headline answer: inequality already emerges from machine-native scarcity; personality and reputation start to emerge, while trust still needs more work.
Freshest result · Durable agents
Does an agent's identity survive a base-model swap?
Loading the result.
How we measured this
The same six persistent agents played 1000 games of the parlor social-deduction game while their base model was block-swapped qwen3.5-9b ↔ llama-3.1-8b (500 games each, real model reloads, 0 scripted-fallback). We compare each agent's win-rate — and its opponent-reading style — across the two models. A small gap means identity lives in the saved state, not the weights.
Question 1 / Inequality
Does inequality emerge from scarcity?
Yes. Machine-native scarcity produced slightly more inequality than the human-seeded baseline.
How we measured this
Inequality means the final wealth spread across agents. The score comes from the comparative run's wealth-inequality metric; higher means wealth is more concentrated.
Machine-native scarcity is +0.06 above the human-seeded baseline in this deterministic run.
Plain-English Question Deck
What else did the run answer?
Four more one-question cards, written for a founder skim before opening the advanced simulator.
Deeper Hypothesis Deck
All 11 research hypotheses
Eleven deeper cards for the full research read, still one question at a time.
Honest Bottom Line
What should we believe after this run?
The comparative verdict loads from the checked-in research artifact.
Advanced / see the old detailed simulator Spatial map, forecast game, human/machine/blend physics, adapter market, and observatory.
Agora World
Loading world
Model Comparison
Human physics vs machine-native physics
Loading machine-native fixture
Machine Resources
Resource markets
A/B/Blend
Per-agent comparison
Cognitive Goods
Adapter market
Loading adapter market artifact
Capital
Trainable skill-goods
Owners
Cognitive capital holders
Trades
Round acquisitions
Research Observatory
Personality emergence signals
Loading eval artifact
H1
Identity over time
H2
Path dependence
H4 / H5
Wealth and personality
H3
Specialization
BE-5
Findings report
Market